Wayne -

The point I was making, but clearly not well, is that when the median survival differences are generately 3% or less, using primarily whole numbers for discussing medians loses precision. Sort of like using a mercury thermometer to evaluate difference in temperature of 1 degree. In the Nature Genetics paper, there are 16 drugs shown in Supplemetary Table 4 with % change in median survival shown. There, Minocycline shows -2.3%. There are no P values given for anything other than ALS TDI 846, as I recall. Just as you found, I've noticed that using the numbers in the Research Archive result in similar numbers, but the difference leads me to the belief that ALS TDI has more precise numbers. I am not trying to make a big issue of this, only commenting that it seems the summary table has more precise numbers, so I use those values instead. It does look like you found a mistake.

Regarding the P value, I just rounded the number you mention to 0.785.

While they likely have more data, For someone like me using the data available to me, I am simply challenging the conclusion of worsening progression when treating pre-symptomatically. And, if there exists a better study where there was statistical significance, then I wonder why that is not used instead?

More generally, I am led to believe that ALS TDI has never had statistical significance for a difference in median survival in any study except for ALS TDI 00846. And, so it seems wrong (based on information available to me), that they can conclude that any drug was detrimental, just as they cannot conclude that any other than 00846 was beneficial in the SOD1 mouse.

Another hypothesis: Of the 15 other drugs shown in Supplementary Table 4 of the Nature Genetics article regarding ALS TDI 00846, many of the median survival differences would become statistically significant, and thus valid for judgments, if the sample sizes were doubled.

When data is not statistically significant, by the author's definition, it does not make sense to then make judgements about changes in median survival. By the self-imposed definition, the control vs. treatment were not different.

---

per·se·vere [pur-suh-veer] - to persist in anything undertaken; maintain a purpose in spite of difficulty, obstacles, or discouragement; continue steadfastly.

The table I am referring to has already done the math. It lists only % change in median survival. It gives no value of days survival for treatment or control.

That table can be found on page 4 of the PDF file here: http://www.nature.com/ng/journal/v42/n5/extref/ng.557-S1.pdf

If you look back at my summary of the control group median survivals, you'll sare they are whole numbers, with one exception ending in".5".

With an average median survival of 130 days for the control, and as you mentioned, the denominator of a percent change calculation, using only whole numbers results in discrete steps of 0.8%. The is a HUGE step when considering all 15 numbers are 3.2 or below (+ and -). This would mean that the precision is only 25% of the max effect, and would be pretty poor. So, I hope that whole numbers of days are not used, and someone tracks it with more precision. But, this may not be feasible as the evaluation probably is done once per day.

---

per·se·vere [pur-suh-veer] - to persist in anything undertaken; maintain a purpose in spite of difficulty, obstacles, or discouragement; continue steadfastly.

There is a much bigger error for Ceftriaxone there. It shows 0.004, but the correct value is 0.4 based on the data in the Research Archive.

But, again, neither one matter, since by statistical significance, the median change is not "real" anyhow. In fact, it is probably misleading to include that table without P values or a comment of lack of statistical significance. Without a disclaimer, readers are (wrongly) invited to compare the values, in my opinion. (However, I will admit that I am not familiar with the Jackknife Outlier Distance Method, and realize that the "distance" column is probably somewhat a surrogate for P value)

---

per·se·vere [pur-suh-veer] - to persist in anything undertaken; maintain a purpose in spite of difficulty, obstacles, or discouragement; continue steadfastly.

Thinking about the SUPER accelerated SOD1 mouse model in comparison to humans with ALS, Nemesis' comments, and the imprecision of daily evaluations with historical differences in median survivals of a couple to few days.....

We know that the metabolism of mice is much faster than humans. Yet, we seem to use very similar dosing intervals (e.g. Once per day) and the same doses per body mass. In fact we may even use MORE frequent dosing intervals in humans (e.g. Ceftriaxone twice daily, Riluzole twice daily). So, as a NON-biologist, who didn't particularly enjoy AP Biology 25 years ago and admits I am out of my knowledge zone, I wonder how this may play a role in test results in terms of half lives in mice versus half lives in men? My uneducated intuition is that each mouse would have much greater fluctuation in drug effect between each dose than humans. If true, could this ever result in less efficacy in SOD1 mice than may be possible in humans (i.e. false negatives)?

Dr. Vieira? Or other forum members with a better grasp on Biology than me?

---

per·se·vere [pur-suh-veer] - to persist in anything undertaken; maintain a purpose in spite of difficulty, obstacles, or discouragement; continue steadfastly.

In an attempt to satisfy my own curiosity regarding mouse dosing versus human dosing, I came across this article:

http://www.fasebj.org/content/22/3/659.full

This article does not address dosing frequency, but suggests that dosage per body mass should not be equivalent between mice and humans. While this makes sense, this has me a little confused in reading of the doses of NP001 in human clinical trials vs. mouse studies, where the doses seemed the "same" on a mg/kg basis. The article above calculates a dosage 12.3 times lower for humans versus mice on a mg/kg basis.

For the NP001 real world example, see here (poster SW294 beginning on page 49): http://www.mndassociation.org/document.rm?id=2149 Human dosing is discussed in the Methods section, while mouse dosing is described in Discussion section (all on right half of pages).

Dr. Vieira - Will you please let us know the assumed dosing correlation made at ALS TDI?

---

per·se·vere [pur-suh-veer] - to persist in anything undertaken; maintain a purpose in spite of difficulty, obstacles, or discouragement; continue steadfastly.

A friend reminded me of the very clever idea that has been presented by ALS TDI. Among other places, it was presented as Poster P41 from the International Symposium on ALS/MND (page 82 as printed in the document):

http://www.mndassociation.org/document.rm?id=2139

It is an infrared motion sensing system to continuously monitor disease progression as correlated to reduced movement count in mice. I think this would definitely address the precision issue with neurological scores and possibly even the endpoint, if correlated to a low threshold number of movements in a given time period. This could replace currently very discrete measurements with continuous ones. I expect this will improve study power.

---

per·se·vere [pur-suh-veer] - to persist in anything undertaken; maintain a purpose in spite of difficulty, obstacles, or discouragement; continue steadfastly.

It also reduces handling of the mice, and handling may affect disease progression. TDI is doing all in its power to speed up and improve the pre-clinical process.

On apocynin details:

Apocynin study design and statistics:
52 treated and 52 control mice
100% litter matched
27 Females and 25 males in each cohort

Median Age at Onset of Paresis (days):
Male Control: 107
Male Treatment : 113
Female Control: 117
Female Treatment: 113
Overall Control: 112
Overall Treatment: 113

Male difference with drug: +6 (p=0.009 by Wilcoxon)
Female difference with drug: -4 (p = 0.138 by Wilcoxon)
Overall difference with drug: +1 (p = 0.654 by Wilcoxon)


Median Age at ALS Related Death (days):
Male Control: 123
Male Treatment : 126
Female Control: 128
Female Treatment: 129
Overall Control: 125
Overall Treatment: 128

Male difference with drug: +3 (p=0.085 by Wilcoxon)
Female difference with drug: +1 (p = 0.905 by Wilcoxon)
Overall difference with drug: +3 (p = 0.267 by Wilcoxon)


More later on metabolism of drugs in mice

---

Fernando G. Vieira, M.D.
Director of In Vivo Operations
ALS Therapy Development Institute

Thank you, Dr. Vieira.

Another way of looking at the data:

Treatment delay of onset: M= +6, F= -4. (Note: onset of controls 107 and 117 days respectively)

Treatment period of decline: M= -3, F= +4. (Note: controls decline 16 and 11 days respectively)

The data "suggest" that the longer you delay onset, whether by apocynin treatment or by being female, the quicker the decline once it begins. And that if you accelerate onset (in females) with apocynin, the period of decline is longer. Also note that the treatment effect as a percentage of time to onset is rather small, but it is substantial with respect to duration of decline. Duration of decline data might produce more robust statistical confidence than the days-to-onset and the days-to-death data.

It is somewhat but not entirely counterintuitive that a treatment which delays onset would "seemingly" accelerate decline. Several theories could account for this:

1. The treatment impacts processes which are directly related to onset, but not directly related to death, the latter processes already being underway long before onset.

2. The treatment has opposite effects prior to onset of the full-scale degenerative cascade, and after its onset.

3. The treatment's effect has more to do with impact on hormones and their receptors, making males "more female" and females "more male", than on the neurodegenerative disease process itself.

4. Certain forum denizens can probably plug in their own theories.

5. All these numbers bouncing around several days this way and that in a pattern that seemingly doesn't make sense, is in fact meaningless, and the statistical levels of confidence which are being applied to it are a combination of correlation shopping and/or failure to account for perturbing influences on the experiment and the assumptions regarding statistical variability that lurk behind the experiment.

* * * * * *

The results looked a lot more interesting when they were graphed as days-to-death each mouse, comparing the control distribution to the treatment distribution. Yep, can see it real easy with your naked eye. But what seems obvious to the naked eye ain't necessarily so.

In the broader perspective, we've got this apocynin mouse trial probably at least as good as any, and when you start to ask difficult questions about what the data mean in relation to the treatment, the whole thing falls apart. Perhaps if different questions were being asked in the beginning, the data produced could have produced something more informative (say for instance, revealing mechanisms of onset, or differences between males and females), but of course for starters the question everyone wanted answered was whether it would help the mice live longer with ALSTDI and not some other lab doing the testing. .........Meanwhile, it was apocynin, something we patients can't get our hands on (as opposed to Picrorrhiza which contains kutkins etc. with their own pharmaceutical actions), and nobody really knows what high-copy G93A mouse data means in relation to human ALS anyhow.

Standardized Picrorrhiza as part of a fairly comprehensive c**ktail in the '43 mouse would be more interesting, despite the fact that's a poorly standardized mouse model. At least it probably models much more closely what happens in humans. And the poor calibration of the model is not a showstopper if you're applying "take your best shot" therapeutic regimes intended to halt disease progression in the mouse, rather than "trialing a drug".

To put it another way, if you're trialing a treatment in the 93 mouse, and you get P= .05, even though you're pretty sure (although perhaps you shouldn't be) that the effect was real in the mouse, the probability that it's gonna help humans is fairly low esp. if it's not a c**ktail. So success in the mouse is probably failure where it counts-- in human beings.

But if you're trialing a c**ktail treatment in the 43 mouse, and you get P= 0.2, there's a good chance you can replicate, and there's a good chance that what you've got is relevant to human therapeutics. So although it may not be very impressive, it's something that may well lead to real benefit to ALS patients if followed up on.

* * * * * * *

After this closer look at the apocynin data, I've pretty much lost interest in it. But more than that, it's a reminder how little relevance trials in the high-copy G93A mouse have to development of therapeutics for human beings. The very things that make it attractive as a lab beastie are the things that cast a cloud of doubt over its usefulness as a model for development of therapeutics for human beings.

--Dave J.

Over the last three years, a lot has changed in perspective among the plugged-in ALS community (i.e., this forum). The value of placebo-controlled trials, the relevancy of the Mighty '93 Mouse, the emergence of other animal models, the distinction between onset processes and decline processes, the complexity of decline processes, the practical need for therapeutic development to focus on post-diagnosis decline processes, the need for therapeutics to be c**cktails on that account, the discovery that noisy patients can be used as resources rather than simply dismissed as the enemy of science, ...... I probably left a bunch of stuff out.

In order for any of this to matter to ALS patients, the medical profession (establishment or otherwise) is going to have to produce effective therapeutics. Otherwise everything is just wasting money and lives. And therein lies a problem: ALSTDI's mission is biotech, not human clinical trials or commercialization of therapeutics.

ALSTDI has some sense of how the ground has shifted over the last 3 years. I suspect that most of the rest of the ALS industry is somewhere between 95 and 100% clueless. Despite vehement critique here, Packard Center has probably not yet figured out that the ceftriaxone trial is the perfect poster child for clinical trial reform since it even has a smoking gun. MDA dropped a hint last year that the concept of historical controls (as opposed to placebo) in human trials has merit, but since this was in the context of yet several more lithium trials, maybe they were just trying to look good after the patient led lithium trial and hope the whole thing will be forgotten since ALS patients tend to die off pretty fast.

No matter how good ALSTDI gets at biotech, if the path from ALSTDI's labs to the clinic is blocked by damnfool ignorance on the part of the rest of the industry, everything ALSTDI does is for naught. It's not just ALS patients who are ripped off, the drug companies themselves are ripped off.

Therefore in my opinion, in order for ALSTDI's work to be more effective, it's not just a matter of doing mousework etc. better, somebody's gotta educate the rest of the pipeline what they need to do in order to stop constipating it, in their own interests. Heck, drug company investors don't understand any of this stuff, their response to the pipeline constipation is to sink huge amounts of money into advertising whatever it is they've got, and to keep politicians bought off. The notion that maybe there's something fundamentally wrong with the way research is done hasn't crossed their minds. The drug researchers in their own companies aren't gonna tell the investors, and probably don't even "get it" themselves being so stuck in the box of past habit.

There's a huge educational problem. Patients can't do the education because nobody listens to us, we're the people who are supposed to shut up and die quietly. Maybe the ALS industry will listen to ALSTDI. I think that education of the pipeline needs to be a key part of ALSTDI's business plan, otherwise the pipeline remains constipated and everything ALSTDI does has no place to go other than literally down the rathole.

The task is probably not as huge as it first might seem. ALSTDI can't educate everyone and doesn't need to. I doubt any of this matters to MDA or the patient advocacy organizations, their needs are not served by effective therapeutics. So don't waste there. I suspect there are potential pipeline partners out there who are open-minded to change and want a piece of real action, and the education is probably a matter of identifying those potential pipeline partners and bringing them up to speed.

--Dave J.